Maternal age is highly associated with non‐chromosomal congenital anomalies: Analysis of a population‐based case‐control database

The role of maternal age in the development of non‐chromosomal congenital anomalies (NCAs) is under debate. Therefore, the primary aim of this study was to identify the age groups at risk for NCAs. The secondary aim was to perform a detailed analysis of the relative frequency of various anomalies.


| I N TRODUC TION
Worldwide, 3-5% of births are affected by a congenital anomaly (CA), 1 representing a leading cause of infant mortality. 2 Based on the EUROCAT survey, the average European relative frequency of birth defects is 23.9 per 1000 births. 3 According to the 2010 Global Burden of Disease study, congenital anomalies (CAs) are responsible for 6% of infant deaths worldwide, 4 whereas other studies suggest that around 20% of neonatal and infant mortality is caused by CAs. 5,6 CAs have a significant burden on society as a whole, namely, the affected families and both health and social care systems. In addition, CA-related hospitalisations are extremely costly, accounting for 3.0% of total hospitalisations and 5.2% of total hospital costs, with an estimated annual cost of $22.9 billion in the USA in 2013. 7 These facts highlight the importance of CAs globally for healthcare systems, research, prevention and screening. Appropriate intervention must be considered a public health priority.
For several known maternal lifestyle risk factors and chronic illnesses, there is a clear association with the occurrence of CAs. For example, according to a meta-analysis, maternal smoking during pregnancy increases the odds of CAs (odds ratio [OR] = 1.18, 95% CI 1.03-1.34). 8 The riskincreasing effect of maternal diabetes is also considered in genetic screening. The effect of pre-gestational diabetes is indeed pronounced according to a meta-analysis (relative risk [RR] = 2.66, 95% CI 2.04-3.47). 9 There is also a well-known correlation between maternal age and chromosomal anomalies, but we have much less information about maternal age as a risk factor in the case of non-chromosomal congenital anomalies (NCAs). The significant role of maternal age in their development is probably established, but the exact details are still the subject of active research. In addition, age distributions of NCAs in the literature are inconsistent. Some studies show a riskincreasing effect either only for the young 10 (generally defined as under 20 years) or only for the advanced 11 (generally defined 35 years or more); others show an effect for both age groups.
We aimed to identify maternal age-related risk groups without arbitrary age categories and to focus on screening options based on maternal age -an approach which is currently missing in the protocols for NCAs. Our hypothesis was that very young (expected to be <20 years) and advanced maternal age (expected to be ≥35 years) increase the risk of NCAs.

| Study design
Our population-based study investigated the relative frequency of CAs in relation to maternal age over a period of almost 30 years in Hungary. This study obtained cases from the Hungarian Case-Control Surveillance of Congenital Abnormalities (HCCSCA) and the total number of live births during the study period from the Central Statistical Office (KSH).
We identified high-and low-risk maternal age groups using the restricted cubic spline model instead of comparing arbitrary age categories. 12 We report our population-based study according to the recommendations of the STROBE guideline (Table S1). 13

| Setting
Our study is an analysis of the HCCSCA (established in 1980, and terminated in 2009). 14 Data collection was changed in 1997 (affecting only the collection of matched controls that were not used in the current study), slightly modifying the structure of the HCCSCA. Data collected through the HCCSCA between 1980 and 2009 were unified into one large validated database. 15 In 2002, after one mother's objections, the legal background of data privacy was questioned, and data collection was suspended until 2005.
Since 1962, reporting patients as cases with CA to the Hungarian Congenital Abnormality Registry (HCAR) has been obligatory for physicians in Hungary, from birth until the end of the first postnatal year. The HCAR was founded in 1962 as the first national-based registry of CAs globally. 16 Since 1984, the prenatal diagnostic centres have also been asked to report malformed fetuses diagnosed prenatally with or without elective termination of pregnancy to the HCAR. Cases have been enrolled in the HCCSCA from the HCAR since 1980.

| Ethics and patient consent
Ethics approval for data analysis was obtained from the Scientific and Research Ethics Committee of the Medical Research Council, Hungary (BMEÜ/920-3/2022/EKU). There are no identifiable registry data reported in our study. National legislation does not require informed consent to register a baby with a congenital anomaly. 17 Patients were not involved in the design and conduct of this research.

| Participants
Cases with CAs in the HCAR were enrolled to the HCCSCA if they met all the following selection criteria: (1) reported to the HCAR within 3 months after birth or elective termination of pregnancy, (2) none of the three mild CAs (hip dislocation, congenital inguinal hernia and large haemangioma) was present alone, and (3) CA syndromes were not caused by gene mutations or chromosomal anomalies with preconceptional origin. In our analysis, we excluded cases with incomplete data or the co-presence of chromosomal anomalies ( Figure 1). The main task of the HCCSCA has been the detection of teratogenic/ fetotoxic agents and other environmental effects during pregnancy resulting in the development of birth defects.
The case group contains live births, stillbirths and elective terminations of pregnancies following prenatal malformation diagnosis. For the total number of cases and controls, the total number of live births by maternal age in Hungary during the study period was obtained from the hungarian Central Statistical Office (KSH).

| Variables and data sources
The following information about every patient was recorded during data collection: NCA(s), gender, maternal age, paternal age, birth date, birthweight, gestational age, area of mother's residence, birth order, mother's and father's qualifications, employment status and type of employment, mother's marital status, outcome of previous pregnancies, maternal diseases during pregnancy (by month of pregnancy), medication during pregnancy (by month of pregnancy), and the mother's smoking habits and alcohol consumption patterns. 15 Maternal age was recorded at the time of delivery or termination of pregnancy due to fetal anomaly.
Data on maternal diseases, lifestyle factors and medication during pregnancy were collected in multiple ways. Mothers provided all their medical documentation about their ongoing pregnancy, and professionals recorded it (prospective, medically recorded data). A questionnaire was then mailed to the mothers containing questions about maternal diseases, pregnancy-related drug treatments and pregnancy supplements (retrospective, maternal self-reported information). Lastly, regional nurses visited all mothers. The nurses helped mothers collect and present their medical records and answer the questionnaire (Table S1).
We performed our analysis by disease categories as defined by the International Classification of Diseases (ICD)-10, which ensures an accurate categorisation. Even though the definition of certain anomalies may have changed during the study period, their ICD categorisation at the level of our analysis remained consistent.

| Bias and evidence synthesis
The maternal ages were recorded based on birth certificates, ensuring a very high level of data accuracy. The unique nature of data collection and verification further enhances the reliability of the data. However, the classification of outcomes was not consistent over the long study period. When converting different ICD categories to each other, the groupings used do not always match with complete accuracy.
We used the GRADEPRO tool to assess the level of evidence for our results. 18

| Statistical methods
Primary data extraction and organisation were carried out in Microsoft EXCEL. Statistical analysis was carried out in R (v4.1.3). 19

F I G U R E 1 (A) Study plan. (B)
Age distribution of cases and total population by age.
The aim of our analysis was to determine the high-risk maternal age for each non-chromosomal anomaly (NCA) category. We used a two-way approach.
First, we identified the best 10-year period of maternal age corresponding to the anomaly's lowest relative frequency. Risk was calculated as: number of cases among live births + stillbirths + elective terminations of pregnancies following prenatal diagnosis of malformation/total number of live births in the population. Risk ratios for each year were determined by taking the best 10-year period as a 'reference risk'. Note that, despite the case-control approach, RR could be used because data collection included the whole population). Cases with a maternal age <13 (one case) and >45 years (nine cases) were excluded because the very low number of cases in these maternal age ranges would have made the regression unreliable. The confidence interval of relative frequency was estimated according to Agresti & Coull. 20 Secondly, we fitted a non-linear, non-parametric logistic regression model on the original data (namely, a restricted cubic splines model using five knots at the 0.05, 0.275, 0.5, 0.725 and .95 quantiles, as recommended in the literature; explanatory variable: maternal age; response variable: presence or absence of NCAs) using the 'rms' R package (v6.2.0). 21 The resulting relative frequency estimates of the regression were transformed to the RR scale to enable graphical representation in the figure showing the year-by-year risk estimates calculated above.
A grouping of NCA categories based on high-risk maternal age was done by considering the confidence bands in addition to assessing the shape of the curves: a curve may appear U-shaped at first glimpse but the risk increase is not necessarily statistically substantiated in both directions, i.e. the confidence band may contain the RR = 1 line corresponding to zero effect.
All confidence intervals were calculated at a confidence level (1-α) of 95%.

| Participants
A total of 31 128 cases with NCAs were identified in Hungary (Table S2); during the study period there were 2 808 345 live births in the country (Figure 1). Figure 1 presents the age distribution of the study population, showing that 7.66% of births fell into the very young (≤19) and 6.62% into the advanced (≥35) maternal age categories. Within this group, 1.11% of births represented mothers over 40. This means that 14.28% of births were in the maternal age groups expected to pose an increased risk.
Mean maternal age was practically the same among cases (26.0 years; SD = 5.4) and in the reference population (26.1 years; SD = 5.1).

| Descriptive data
Thanks to the population-wide data collection, we had individual information about the cases. In the table below, we have summarized some of this information ( Table 1). The most notable is the sex of the fetuses, which is around 65% male.

| Outcome data
The relative frequency of NCAs in the study period was 1.1% after excluding cases with only mild anomalies and chromosomal anomalies.

| The risk-increasing effect of advanced and very young maternal age
In the first step, all NCAs were analysed together ( Figure 2 advanced maternal age increase risk even more. Even though the confidence range becomes wider in the very young and old maternal age groups due to the low number of cases, the trend is still clear. In the next step, NCAs were analysed by ICD category (Figure 3). In the case of certain ICD categories, both lower and higher maternal ages exert a risk-increasing effect, namely: circulatory system (Q20-Q28), cleft lip and palate (Q35-Q37), genital organ system (Q50-Q56), musculoskeletal system (Q65-Q79) and digestive system (Q38-Q45). A U-shaped regression curve can describe the relation. Observing the regression line in the case of musculoskeletal and digestive system anomalies, there is an increased risk of birth defects in very young mothers. In cases of circulatory system anomalies and cleft lip/palate, the increased risk is more pronounced for advanced age mothers. There is no expressed difference in the risk-increasing effect when comparing the lower and higher maternal age ranges in the case of the genital organ system.

| The risk-increasing effect of advanced maternal age only
In the case of CAs of the urinary system (Q60-Q64) and malformations of the eye, ear, face and neck (Q10-Q18), advanced maternal age exerts a risk-increasing effect while young age does not. However, looking at the figure regarding the malformations of the eye, ear, face and neck (Q10-Q18), the results are somewhat inconsistent, and the increase in risk becomes clearly significant only >40 years.

| The risk-increasing effect of very young maternal age only
The nervous system anomalies (Q00-Q07) category is the only one where only young maternal age is associated with increased risk. Looking at the entire low-age group, the risk increase is 25%. For the very young (<20), there is an apparent increase in risk that is even higher.

| Congenital anomalies not related to maternal age
According to our analysis, respiratory system anomalies (Q30-Q34) could not be proven to be associated with maternal age.

| Level of evidence
When all NCAs were analysed according to maternal age, the young and advanced age groups were found to have moderate certainty of NCAs ( Figure S1).

| DISCUS SION
The main findings of this study confirm our hypothesis. The relative frequency of NCAs strongly depends on maternal age. Our data are of clinical importance because, based on these results, preventive and screening interventions can be applied according to maternal age groups. Furthermore, our research shows that very young and advanced maternal age increase risk when all NCAs are examined together. This finding is particularly important, considering that chromosomal anomalies (with a well-known correlation to maternal age) were excluded from the analysis.
Although the topic has been eagerly investigated, age distributions of various NCAs are inconsistent in the literature. In line with our findings, Reefhuis et al. 22 showed that women <20 and ≥35 years are at increased risk of having a fetus with an NCA. Croen et al. 23 also found this association in their data analysis from the California Birth Defects Monitoring Program, excluding the Afro-American population. Looking at all the NCA categories combined, other studies have shown a risk-increasing effect of older maternal age. 11 This may be due to the riskincreasing effect of chromosomal anomalies that occur more frequently with advanced maternal age. Hollier et al. 11 suggest that the accumulation of environmental exposures over time may also have a risk-increasing effect.

F I G U R E 2
Analysis of all NCAs by maternal age. The figure shows the estimated risk ratios of NCAs as a function of maternal age with the best 10-year period as 'reference risk' (circle markers). The best 10-yearperiod is highlighted in light grey. The black curve shows the result of the restricted cubic splines regression; the dark grey area is its confidence range.
Analysing data from the EUROCAT database, Loane et al. 10 point out that more emphasis should be placed on screening very young mothers, who are more likely to have several risk factors. Zhang et al. found an increase in risk for extremely young mothers and mothers younger than 25 years. The authors emphasise that statistically significant differences in NCA relative frequency were found between different levels of maternal education.
Based on the current guidelines, there is no recommendation for screening NCAs with regard to maternal age. [24][25][26] Maternal age has previously been shown to be a relevant risk factor for chromosomal anomalies, and this pressure for age-based screening has significantly increased the detection rate.
Neural tube defects (NTDs) and congenital heart defects (CHDs) should be discussed separately, as fetal Looking at the NTDs (ICD-10 Q00-Q07) together, we observed an increased risk in very young mothers. The literature is not consistent on the age effect. Most studies have suggested a 'U-shaped' association between maternal age and the relative frequency of NTDs. 27,28 Other groups suggest that a higher risk of NTD is likely associated with increased maternal age. 29 The heterogeneous results may be the consequence of inappropriate NTD definition, as grouping was not uniformly applied across studies. Some anomalies were explicitly associated with young maternal age (e.g. anencephaly), 30,31 whereas other isolated anomalies were more common with older maternal age (e.g. spina bifida, encephalocele). 31 From a clinical point of view, finding a clear association between maternal age and the relative frequency of CHDs (ICD-10 Q20-Q28) is an important task. Currently, there is no maternal age-related indication for fetal echocardiography. 32,33 This finding is of particular significance because the effects of chromosomal anomalies did not modify the relative frequency found in our study. Various studies have been published about the risk-increasing effect of older maternal age, but it is important to note that the co-occurrence of chromosomal anomalies is most significant at this age. 34 As there is no additional screening opportunity for the other NCA groups, age-adjusted ultrasound examinations in these age groups must focus on these organ systems.
Particular attention should be paid to more frequent differences in the low or high maternal age categories. Examining what may be behind the risk-increasing effects of each age group can help identify the right prevention options. The teratogenic effects associated with the lifestyle of mothers becoming pregnant at a very young age and the lack of primary prevention options may largely explain the vulnerability of this age group, including smoking, drug and alcohol abuse (substance abuse together 41.0%), lower social status, lower educational attainment 35 and the lack of adequate folic acid supplementation typical of conscious childbearing. 36 This investigation of socioeconomic differences in the use of supplements found inequalities that benefit the wealthier and more highly educated white mothers. The lack of folic acid intake is clearly associated with a higher risk of NTDs. 37 In contrast, it is worth investigating the possible correlations between maternal chronic diseases and conditions relating to the risk-increasing effect of advanced maternal age. The agerelated decline in oocyte quality and deteriorating repair processes could be the subject of basic research regarding CAs of the urinary system and facial malformations.
Our results suggest that incorporating the age aspect into screening protocols can increase the possibility of early detection of NCAs. Although the present study is not sufficient to confirm an isolated evidence of age effects, and the influence of lifestyle factors typically associated with age categories may be significant, and age alone may represent a well-defined, clear risk factor.

| Strengths and limitations
The strengths of our analysis are the large number of cases, the unique database and data collection methods, and that the collection of data on maternal age is highly accurate. In addition, the novel statistical approach employed may better reflect reality without using arbitrary groups.
Considering the limitations of this work, slight changes in screening methods and detection rates during the long study period could be mentioned. In addition, the definitions of some individual anomalies differed between the years or were even missing. These documents were also structured for the ICD-10 categories; however, it was impossible to identify them precisely in some cases. Finally, the main limitation is the lack of a multivariate model. However, this stemmed from the nature of the population-based study.
The generalisation of our result is substantiated, as the enrolled patients represent the entire selected geographical region.

| Implications for practice and research
Based on our results, we suggest maternal age-based screening for CHDs.
Further prospective data collection is needed to assess the problem more accurately and to consider confounders. For example, an international congenital anomaly registry that collects all pregnancy data prospectively and allows multivariate analysis or observational clinical research with longer follow-up periods might give additional insight into this topic. In addition, screening protocol modifications require further health-economic studies, but the risk-increasing effect of maternal age can already be considered for individual cases using our results.

| CONCLUSION
Our results show that certain NCAs are strongly associated with maternal age: a clear increase in risk can be observed for very young or advanced maternal age, or both -exact age limits varying by disease. Taking this into consideration, improved screening protocols should be implemented. Current protocols do not include maternal age-based recommendations for either fetal echocardiography or fetal neurosonography, which would be useful in detecting the respective NCAs. Moreover, in addition to mothers of advanced age, due attention should also be paid to very young groups.

AU T HOR C ON T R I BU T ION S
BP: conceptualisation, project administration, writingoriginal draft. ÁM: conceptualisation, writing -review & editing. GA: conceptualisation, formal analysis; visualisation. DSV: conceptualisation, formal analysis, visualisation. AH: conceptualisation, formal analysis, visualisation. SV: conceptualisation, methodology, writing -review & editing. FB: conceptualisation, writing -review & editing. PH: conceptualisation, writing -review & editing. NÁ: conceptualisation, supervision, writing -original draft. All authors certify that they have participated sufficiently in the work to take public responsibility for the content, including participation in the paper's concept, design, analysis, writing and revision.

AC K NO W L E D GE M E N T S
None to declare.

C ON F L IC T OF I N T E R E S T S TAT E M E N T
None declared. Completed disclosure of interest forms are available to view online as supporting information.

DATA AVA I L A BI L I T Y S TAT E M E N T
All data were included in the manuscript.

F U N DI NG S TAT E M E N T
Sponsors had no role in the design, data collection, analysis, interpretation or article preparation.

E T H IC A L A PPROVA L
In our study, cases were obtained from the Hungarian Case-Control Surveillance of Congenital Anomalies (HCCSCA) and the control group comprised live births from the Central Statistical Office (KSH) during the study period. Ethics approval for data analysis was obtained from the Scientific and Research Ethics Committee of the Medical Research Council, Hungary (BMEÜ/920-3 /2022/EKU). The study was conducted in accordance with the Declaration of Helsinki. Patients were not involved in the design or conduct of this research.